Share your thoughts, 1 month free Claude Pro on usSee more

Visual Search on V*

92.67Accuracy

Dynamo + RL

Updated 25d ago

Evaluation Results

Method
Dynamo + RL 2026.06	92.67	-	-	-	-	-	-	-	-
DeepEyes 2025.09	90.1	-	-	53.72	-	-	-	-	-
Chain-of-Focus 2025.09	88	-	-	50.94	-	-	-	-	-
ViGoRL-7B 2025.09	86.4	-	-	50.53	-	-	-	-	-
Dynamo 2026.06	85.86	-	-	-	-	-	-	-	-
Visual Para-Thinker++ 2026.06	85.7	-	-	-	-	-	-	-	-
Dynamo + RL 2026.06	85.34	-	-	-	-	-	-	-	-
RL 2026.06	84.82	-	-	-	-	-	-	-	-
Dynamo 2026.06	84.29	-	-	-	-	-	-	-	-
InterSketch 2026.05	83.8	-	-	-	-	-	-	-	-
AdaVaR-7B 2025.09	83.4	-	-	55.82	-	-	-	-	-
MiMo-Embodied 2025.11	82.72	-	-	-	-	-	-	-	-
Thyme-7B 2026.05	82.2	-	-	-	-	-	-	-	-
Visual Para-Thinker 2026.06	82.2	-	-	-	-	-	-	-	-
DeepEyeV2-7B 2026.05	81.8	-	-	-	-	-	-	-	-
MiMo-VL 2025.11	81.68	-	-	-	-	-	-	-	-
Visual Para-Thinker++ 2026.06	80	-	-	-	-	-	-	-	-
Direct 2026.06	79.06	-	-	-	-	-	-	-	-
GPT5-mini 2026.06	78.6	-	-	-	-	-	-	-	-
VLAA-Thinker-7B 2025.09	78.5	-	-	54.2	-	-	-	-	-
GRPO 2026.06	78.4	-	-	-	-	-	-	-	-
ViGoRL-3B 2025.09	78.1	-	-	45.7	-	-	-	-	-
RL 2026.06	78.01	-	-	-	-	-	-	-	-
Qwen2.5-VL-7B 2025.09	78	-	-	50.9	-	-	-	-	-
Orsta-7B 2025.09	78	-	-	52.04	-	-	-	-	-
AdaVaR-3B 2025.09	77	-	-	50.84	-	-	-	-	-
Qwen2.5-VL 2026.06	76.9	-	-	-	-	-	-	-	-
Gemini2.5-Pro 2026.06	76.7	-	-	-	-	-	-	-	-
Qwen2.5-VL-7B 2026.05	76.4	-	-	-	-	-	-	-	-
Sequential (SFT) 2026.06	76.1	-	-	-	-	-	-	-	-
Qwen2.5-VL 2026.06	75.9	-	-	-	-	-	-	-	-
Visual Para-Thinker 2026.06	75.9	-	-	-	-	-	-	-	-
Majority voting@4 2026.06	75.6	-	-	-	-	-	-	-	-
LLaVA-OV-7B 2025.09	75.4	-	-	-	-	-	-	-	-
LMM-R1 2025.09	74.9	-	-	48.8	-	-	-	-	-
GPT-4o 2025.11	73.9	-	-	-	-	-	-	-	-
GPT-4o 2026.06	73.9	-	-	-	-	-	-	-	-
Qwen2.5-VL 2025.11	73.8	-	-	-	-	-	-	-	-
Qwen2.5-VL-3B 2025.09	73.5	-	-	45.74	-	-	-	-	-
Qwen3-VL-8B 2026.05	73	-	-	-	-	-	-	-	-
InternVL3 2025.11	72.8	-	-	-	-	-	-	-	-
GRIT-3B 2025.09	72.3	-	-	45.8	-	-	-	-	-
Direct 2026.06	71.2	-	-	-	-	-	-	-	-
GRPO 2026.06	70.4	-	-	-	-	-	-	-	-
ThinkMorph-7B 2026.05	67	-	-	-	-	-	-	-	-
GPT-4o 2025.09	66	-	-	53.2	-	-	-	-	-
Sequential (SFT) 2026.06	65	-	-	-	-	-	-	-	-
Majority voting@4 2026.06	64.7	-	-	-	-	-	-	-	-
Qwen2.5-VL 2026.06	63.3	-	-	-	-	-	-	-	-
OVR-7B 2025.09	62.5	-	-	55.76	-	-	-	-	-
MM-Eureka 2025.09	59.7	-	-	52.52	-	-	-	-	-
VLAA-Thinker-3B 2025.09	56	-	-	46.7	-	-	-	-	-
Claude-4-Sonnet 2026.06	45	-	-	-	-	-	-	-	-
GPT-4o 2025.12	-	-	-	66	-	-	-	-	-
SEAL 2025.12	-	-	-	75.4	-	-	-	-	-
Dyfo 2025.12	-	-	-	81.2	-	-	-	-	-
Zoomeye 2025.12	-	93.9	85.5	90.6	-	-	-	-	-
LLaVA-OneVision 2025.12	-	75.7	75	75.4	-	-	-	-	-
Qwen2.5-VL 2025.12	-	79.13	73.68	76.96	-	-	-	-	-
Qwen2.5-VL-CoT 2025.12	-	73.91	76.32	74.86	-	-	-	-	-
Chain-of-Focus 2025.12	-	-	-	88	-	-	-	-	-
Deepeyes 2025.12	-	90.43	84.21	87.96	-	-	-	-	-
Pixel Reasoner 2025.12	-	91.3	77.63	85.86	-	-	-	-	-
IMAgent 2025.12	-	88.7	88.16	88.48	-	-	-	-	-
No fine-tuning 2026.04	-	-	-	-	56	53	60.5	-	-
LVR 2026.04	-	-	-	-	64.9	69.6	60.5	-	-
PEARL 2026.04	-	-	-	-	73.8	82.6	60.5	-	-
PEARL 2026.04	-	-	-	-	69.6	78.3	56.6	-	-
PEARL 2026.04	-	-	-	-	62.8	69.6	52.6	-	-
No fine-tuning 2026.04	-	-	-	-	81.2	86.1	73.7	-	-
PEARL 2026.04	-	-	-	-	81.7	85.2	76.3	-	-
PEARL 2026.04	-	-	-	-	79.6	82.6	75	-	-
PEARL 2026.04	-	-	-	-	75.4	81.7	65.8	-	-
VLM-R1 2026.04	-	-	-	-	72.25	-	-	75.65	67.11
LMM-R1 2026.04	-	-	-	-	49.21	-	-	46.09	53.95
R1-VL 2026.04	-	-	-	-	58.64	-	-	59.13	57.89
Perception-R1 2026.04	-	-	-	-	53.92	-	-	57.39	48.68
Jigsaw-R1 2026.04	-	-	-	-	69.63	-	-	72.17	65.79
Qwen2.5-VL 2026.04	-	-	-	-	60.73	-	-	57.39	65.79
Qwen2.5-VL + GRPO 2026.04	-	-	-	-	80.1	-	-	86.95	69.73
Perceval 2026.04	-	-	-	-	83.25	-	-	90.43	72.37
DeepEyes 2026.04	-	-	-	-	87.43	-	-	91.3	81.58
Pixel-Reasoner 2026.04	-	-	-	-	84.3	-	-	-	-
VL-Rethinker 2026.04	-	-	-	-	56.54	-	-	54.78	59.21
VLAA-Thinker 2026.04	-	-	-	-	47.12	-	-	43.47	52.63
R1-VL 2026.04	-	-	-	-	55.5	-	-	47.83	67.11
OpenVLThinker 2026.04	-	-	-	-	78.01	-	-	76.52	80.26
MM-Eureka 2026.04	-	-	-	-	48.17	-	-	42.61	56.58
Qwen2.5-VL 2026.04	-	-	-	-	62.3	-	-	60.87	64.47
Qwen2.5-VL + GRPO 2026.04	-	-	-	-	84.29	-	-	85.22	82.89
Perceval 2026.04	-	-	-	-	86.39	-	-	86.09	86.84